Completed
Model training progress and high grad norm observation
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Transcription Models Zero to Hero - Data Prep, Train and Serve
Automatically move to the next video in the Classroom when playback concludes
- 1 Introduction and overview of pipeline
- 2 Data requirements: audio recordings and transcripts
- 3 Uploading audio-text pairs and dataset preparation
- 4 Saving word swaps and transcribing with model for better training
- 5 Warning about clean text being out-of-distribution for small models
- 6 Setting up Hugging Face token and Weights & Biases key
- 7 Creating validation set using ChatGPT to rephrase text
- 8 Configuring training settings and advanced parameters
- 9 Baseline evaluation shows 7.09% word error rate
- 10 Training begins with falling loss and word error rate
- 11 Model training progress and high grad norm observation
- 12 Model and logs pushed to Hugging Face Hub
- 13 Inspecting evaluation results and specific corrections
- 14 Spelling improvements and regressions in fine-tuned model
- 15 Deploying model to endpoint with keep warm feature
- 16 Auto-sleep containers and API key access options
- 17 Testing endpoint and transcript download formats
- 18 Evaluation tab features and future text-to-speech plans
